[rehasu]

If you talk to physicists and they tell you we've got something right, stop talking to them.

Every good(!) one knows that we got it wrong. Just a little less wrong than in the previous model. You cannot get it right. It wouldn't be a model but reality as a whole if you got it right. So the only questions are how you can get it less wrong than the last model, and which of the available models is most useful for the problem you are currently tackling. So yes, sometimes you even want to use a more-wrong model because it fits your problem better.

edit: thought about bringing in an example, but honestly it would need too much googling to be precise enough.

[edem]

> sometimes you even want to use a more-wrong model because it fits your problem better.

Or they are equal in fit, but the more wrong is easier to use.